I defended my PhD the 25th of September 2007 at the University of Paris South (Paris 11 - France).
Download my thesis
Download the defense slides


I'm a former member of the Machine Learning/Optimization I&A team in the LRI (CNRS, UMR-8623) under the direction of Michèle Sebag and Nicolas Bredeche.
Our team have operated a fusion with an INRIA one and we actually are the TAO group (mix of INRIA, CNRS and University).

There are my main topics of research (all in a Machine Learning point of view):

• Bayesian Networks
• Optimal Control in difficult cases and Reinforcement Learning:
• I am one of the authors of OpenDP a general and featured framework of reinforcement learning, and in particular of dynamic programming. OpenDP inclues for example major optimization and regression algorithms (including for example the whole weka).
• I am one of the authors of MoGo a top level computer-go program. MoGo uses UCT algorithm for the tree based search, and Monte-Carlo simulations to evaluate the position. MoGo is, since beginning of August 2006, the best ranked go program in the 9x9 Computer Go Server. MoGo has won multiple 9x9, 13x13 and 19x19 computer go international tournaments on KGS. MoGo had many successes, like winning the Gold Medal in the 2007 Computer Game Olympiad for the game of Go, and being the first program to defeat a professional player in 9x9 Go.